Geochimica et Cosmochimica Acta, Vol. 62, No. 18, pp. 3087–3094, 1998
Copyright © 1998 Elsevier Science Ltd
Printed in the USA. All rights reserved
0016-7037/98 $19.00 1 .00
Pergamon
PII S0016-7037(98)00215-4
Isotope-ratio-monitoring of O2 for microanalysis of
in geological materials
18
O/16O and
17
O/16O
EDWARD D. YOUNG,*,1 MARILYN L. FOGEL,2 DOUGLAS RUMBLE III,2 and THOMAS C. HOERING2
1
Department of Earth Sciences, University of Oxford, Oxford OX1 3PR, UK
Geophysical Laboratory, 5251 Broad Branch Road, NW, Washington, D.C. 20015, USA
2
Abstract—Experiments were performed to test the viability of using isotope-ratio-monitoring gas-chromatography mass spectrometry (irm-GCMS) as a means for isotopic analysis of nanomole quantities of O2
released in a vacuum system suitable for laser extraction and fluorination. Several sources of error were
identified and eliminated, including false signals from extraneous scattered ions and adsorption of O2 to metal
surfaces. Results show that with appropriate attention to these potential impediments, coupling high-precision
irm-GCMS to laser sampling of microgram quantities of silicate and oxide minerals is feasible. Copyright ©
1998 Elsevier Science Ltd
ratios determined by irm-GCMS are approximately double
those imparted by current fluctuations known as shot noise
(actual precision ranges from 1.4 to 2.3 times shot noise). These
fluctuations are inherent in the statistics of counting discrete
charges over a fixed time interval and are calculable. Instrument performance relative to shot noise can be used to predict
real-world precision. With this in mind, estimates for the minimum silicate sample size that can be analyzed by irm-GCMS
to a specified precision are obtained using the method of Hayes
et al. (1977) and Merritt and Hayes (1994).
Convention dictates that uncertainties be expressed in terms
of the d notation (McKinney et al., 1950). For oxygen, the d
refers to the per mil (per thousandth) deviation in 18O/16O or
17
O/16O of sample ( x) relative to that of standard (std):
1. INTRODUCTION
Considerable effort has been expended in recent years toward
development of an oxygen isotope microprobe. Laser-based
methods of O extraction combined with gas-source isotope
ratio mass spectrometry (MS) offer the potential for spatial
resolution commensurate with individual minerals in rocks and
analytical precision sufficient for resolving differences in 18O/
16
O of several tenths of one part per thousand. The inability to
measure the isotopic composition of small amounts of O released from mineral samples has been one of the most unremitting problems for those seeking to perform 18O/16O and
17
O/16O microanalysis of silicates and oxides (e.g., Weichert
and Hoefs, 1995; Rumble et al., 1997) and manipulation of
small (e.g., nanomole) quantities of O2 gas with isotope fidelity
is often cited as the limiting factor in laser-based microanalysis
of oxygen isotope ratios by fluorination (Valley et al., 1995).
We investigated the use of isotope-ratio-monitoring gaschromatography mass spectrometry (irm-GCMS) as a means
for measuring 18O/16O and 17O/16O of O2 liberated by laser
ablation and fluorination in nanomole quantities. The irmGCMS method for O2 analysis was investigated independent of
laser extraction and fluorination procedures so that the causes
of any fractionation incurred by the latter could be isolated
from the effects of gas handling and mass spectrometry. Results
show that O2 samples corresponding to microgram quantities of
silicate or oxide mineral material can be analyzed for both 18O/16O
and 17O/16O with precision and accuracy similar to that of conventional dual-inlet MS methods. The irm-GCMS method has
been used to develop a laser-based microprobe for oxygen isotopic
analysis described elsewhere (Young et al., 1998).
d 17O 5 ~ 17R x/ 17R std 2 1! z 10 3
(1)
d 18O 5 ~ 18R x/ 18R std 2 1! z 10 3
(2)
where 17 R x and 17 R std are 17O/16O for the unknown and the
standard, respectively, and 18 R x and 18 R std are the analogous
values for 18O/16O. In practice 17 R and 18 R are represented by
ratios of ion currents for m/z corresponding to the O2 isotopomers 16O16O, 18O16O, 16O18O, 17O16O, 16O17O, and
17 17
O O. The three currents of interest are 32 I, 33 I, and 34 I.
Ratios of 33 I and 34 I relative to 32 I are written as 33 R i and 34 R i
and their relation to the isotope ratios of oxygen is straightforward if there are no interferences from other molecular or
atomic species (Santrock et al., 1985):
R i 5 2 17R
(3)
R i 5 2 18R 1 17R 2.
(4)
33
34
2. THEORETICAL PRECISION
Expressions for uncertainties in measured d values in terms
of the total number of oxygen molecules admitted to the source
are
Determination of 18O/16O and 17O/16O by irm-GCMS
amounts to measuring ratios of ion currents. Therefore, the
maximum precision obtainable is limited by the reproducibility
of ion current detection. Our repeated measurements of reference gases show that realizable uncertainties in oxygen isotope
s
2
d17O
s d218O 5
*Author to whom correspondence should be addressed (ed.
[email protected]).
3087
S
2 3 10 6~1 1 33R i!~1 1 33R i 1 34R i!
33
R i x g«M
5
1
D
2
2 3 10 6~1 1 34R i!~1 1 33R i 1 34R i! 34R i
2
33 2
R
34
Ri 2
x g«M
4
S
D
(5)
3088
E. D. Young et al.
by irm-GCMS declines rapidly for smaller samples. Oxygen
yield for these calculations is the median for most silicates of
geological interest, and mineral stoichiometry has little affect
on the results.
Calculations presented above suggest that irm-GCMS of O2
should, in the absence of other analytical difficulties, enable
high-precision oxygen isotopic analysis of silicate samples with
a maximum three-dimensional spatial resolution of 40 mm.
This conclusion is based solely on the amount of analyte gas,
and so we note that enhancement in spatial resolution is feasible by excavating pits with varied aspect ratio. For example, the
precision predicted for an equidimensional crater of 39 mm
diameter would also apply to a pit with a diameter of 100 mm
and a depth of only 6 mm. Background (blank) oxygen and
other gas-handling factors potentially limit realizable minimum
sample sizes as described below.
Fig. 1. Estimated precision for d18O and d17O as a function of silicate
sample size. s represents shot noise. 4s corresponds to realizable 2s.
d is the diameter of a cylinder of silicate. h is the height of the cylinder.
where x g is the fraction of sample gas that enters the source of
the mass spectrometer, « is the ionization efficiency of the
source expressed as ions per molecule, and M is the total
number of oxygen molecules comprising the sample.
Values of 0.00414 and 0.00075 for 33 R i and 34 R i , respectively, are suitable for most geological materials (e.g., Nier,
1950), whereas ionization efficiency and sample fraction will
vary from instrument to instrument. In the Oxford Earth Sciences stable isotope laboratory, the ionization efficiency for O2
entrained in He has been measured and is 4.2 3 1024. The
fraction of sample that enters the source via the open split is
nominally 0.33. These values are taken to be representative of
what can be obtained with modern instrumentation. Substitution of the measured values for « and x g and incorporation of
Avogadro’s number gives
s d217O 5
0.03199
n O2
s d218O 5
0.00581
n O2
(6)
where n O2 is the amount of sample O2 in nanomoles. Because
actual precision is about two times shot noise, estimates of 2s
precision that can be expected when applying irm-GCMS to the
extraction of O2 from silicate minerals are obtained by multiplying the shot-noise 2sd18O and 2sd17O of Eqn. 6 by a factor of
2 (i.e., actual 2s is equivalent to 4s when s refers to shot
noise).
Figure 1 shows expected real-world precision (2s) in d18O
and d17O prescribed by Eqn. 6 as a function of the amount of
silicate mineral analyzed. The quantity of mineral is expressed
as the dimension of a cylinder of equal depth and diameter, a
shape similar to many pits produced by laser ablation. From the
curves in Fig. 1 it appears that the minimum volume of silicate
that can be analyzed to a 2s precision of 60.2‰ in d18O would
come from a pit measuring 39 mm in diameter, correspond to a
sample weight of approximately 191 nanograms and yield 2.4
nanomoles of O2 gas. The advantage of high precision afforded
3. INSTRUMENTATION
The Oxford University on-line system (Fig. 2) comprises a laserablation fluorination vacuum extraction line evacuated with corrosivegrade turbomolecular and backing rotary pumps, a flow system for
entraining O2 into the He carrier gas, and an irm-GCMS instrument.
The irm-GCMS instrument is composed of a Finnigan MAT 252
gas-ratio mass spectrometer, a Shimadzu GC-R1A gas chromatograph,
and a GC-MS interface. The latter (GC II interface, Finnigan MAT)
consists of an open split, reference gas inlet, and Nafion water-removal
assembly. The open split is described by Merritt et al. (1994). Tank
reference gas injections are obtained by operation of an automated
mixing volume also described by Merritt et al. (1994).
The cryofocusing device, also described as a preconcentration device, for transferring O2 from the evacuated sample loop to flowing He
is shown in Fig. 3. With the exception of the sample loop, for which a
He purged valve is used, carrier gas flow is controlled by fixed lengths
of fused silica capillary tubing acting as flow restrictors and two
pneumatically-controlled on-off valves. Alternation of on-off valves 1
and 2 (Fig. 3) effects reversal of He flow through the cryofocus trap.
Benefits of flow reversal were described by Klemp et al. (1993). The
cryofocus trap consists of a 150 mm length of 1/320 od stainless steel
tube filled with silica gel that was sieved to 2250 mm 1 150 mm and
secured with Ni wool.
A porous layer open tubular (PLOT) fused silica GC separation
column is used in the sample path upstream of the water removal
membrane in order to purify O2. The column has a 0.53 mm internal
diameter, is 10 m in length, and is coated with 5A molecular sieve
adsorbent (manufactured by Chrompak). Operating temperature was
constant at 50°C for all experiments reported here. He flow rate through
the column is 2 mL/min.
4. TEST OXYGEN GASES
All d O and d O values are reported relative to standard mean
ocean water (SMOW). Oxygen from air was adopted as a primary
standard for this study with a nominal value for d18O of 23.5 6 0.3‰
(Craig, 1972). Two tanks of pure O2 gas (grade 5.0) were used as
secondary standards. All standard tank gas d18O values were determined by conventional dual-inlet viscous leak MS methods. Tank 1
18
O/16O was calibrated by quantitative reaction with hot graphite to
form CO2 followed by comparison with the Geophysical Laboratory
CO2 reference gas. The 18O/16O of Tank 2 was obtained by measurements relative to Tank 1 and confirmed by measurement against O2
from air. Tank 2 was used as the Oxford reference gas.
Calibration of 17O/16O was performed by iterative comparison between models for mass-dependent fractionation and measured 18O/16O
and 17O/16O of geological samples. This method is regarded as the most
prudent means for assigning d17O for standard gases (R. N. Clayton,
pers. commun. 1998). Mass-dependent fractionation leads to an array in
a plot of d18O (abscissa) against d17O (ordinate). A mass fractionation
18
17
Isotope-ratio-monitoring of O2
Fig. 2. Partial cutaway diagram of vacuum fluorination line. He plumbing is composed of stainless steel tubing (double lines) or fused silica capillary (single lines). Internal diameter and lengths
are given in mm. Units for other dimensions are as specified.
3089
3090
E. D. Young et al.
the capillary tube to the sample loop of the vacuum system. Pressure in
the glass reservoir was maintained above 140 mbar.
Nominal sample sizes were determined from peak areas by application of the equation
A 5 n O2 ~6.023 3 10 14e x g «R!
32
(8)
where A is the area (V z sec) for the m/z 32 peak, R is the magnitude
of the major-beam collector amplifier feedback resistor (V), n O2 is the
amount of sample gas in nanomoles, and the other terms are defined
above. Accuracy of calibrations based on Eqn. 8 is limited by uncertainties in x g arising from adjustments in He flow rates. Application of
Eqn. 8 for the experiments reported here yields n O2 5 0.62( 32 A). We
note that for irm-GCMS, just as for other MS applications, ionization
efficiency of O2 is about 0.6 that of CO2.
32
5. RESULTS AND DISCUSSION
Results presented below demonstrate that nanomole quantities of O2 gas can be collected in a vacuum extraction system
and analyzed to yield accurate and precise 18O/16O and 17O/
16
O by irm-GCMS.
5.1. Test gases
Fig. 3. Schematic diagram of He flow system for O2 sample trapping
and injection into the irm-GCMS system. Lines represent fused silica or
stainless steel capillary tubing. Circles represent valves. Open valves
are shown with white quadrants adjacent tubing. Closed valves are
shown with black quadrants adjacent tubing. MS marks the fused silica
tube connecting to the mass spectrometer.
array relating terrestrial materials can be described by the equation
(Santrock et al., 1985)
R 5 18R a z k
17
(7)
where k 5 0.04103a 2 0.011387. Kinetic theory (e.g., Bigeleisen,
1965) predicts that the parameter a will depend upon the physicochemical process involved. Values for a are expected to range from 0.500,
representing the asymptotic limit where the effective molecular weights
of the fractionating species containing oxygen are very large, to 0.529
for atomic oxygen. The terrestrial d18O–d17O array is therefore a family
of concave downward curves emanating from the origin and with each
curve defined by a single value for a. As a result, predictions on the
basis of mass fractionation could not be used to define 17O/16O for test
gases directly from their 18O/16O values. Instead, the 17O/16O for Tank
2, the reference gas, was determined by requiring that measured d18O
and d17O for oxygen obtained by laser ablation and fluorination of
silicate minerals (Young et al., 1998) and biogenic phosphates (unpubl.
data of Jones et al., 1998) define an average terrestrial mass fractionation curve with a 5 0.516 (the average value for a is derived by
Matshuhisa et al., 1978). The isotopic compositions of geological
samples and Tank 1 lie on the curve, within error, when d17O for Tank
2 is 20.4 per mil below the curve. Deviations of tank O2 18O/16O and
17
O/16O from the average terrestrial curve have been observed previously (R. N. Clayton, pers. commun. 1998). In order to relate Tank 2
17
O/16O to its 18O/16O with Eqn. 7, the value for a would have to be
0.470. This is lower than the 0.500 limit imposed by simple kinetic
theory, suggesting that mixing between a highly fractionated oxygen (a
byproduct of the manufacturing process) and oxygen with 18O/16O and
17
O/16O closer to terrestrial geological values is the likely explanation
for the offset between Tank 2 and the average terrestrial curve. For
example, 1.45% residuum with d18O of 500.0 per mil and d17O of
232.7 per mil, lying on the average terrestrial curve, mixed with
SMOW oxygen (d18O 5 d17O 5 0), also on the average curve, gives
the isotopic composition of Tank 2.
Aliquotes of Tank 1 and Tank 2 O2 were admitted from a glass
reservoir to the high vacuum system of the on-line apparatus using a
2-m length of 0.06 mm od fused silica capillary tubing (Fig. 2). A
high-vacuum shutoff valve (SMOV, manufactured by SGE) connects
Influences of different segments of the extraction line were
isolated by measuring d18O and d17O after: (1) expansion to the
sample loop with no trapping in the sample loop; (2) expansion
followed by freezing to silica gel in the sample loop; (3)
expansion through the cold trap and fluorine getter of the
vacuum extraction line followed by collection on the sampleloop silica gel; (4) expansion through the extraction line to the
fluorination chamber followed by collection onto the sampleloop silica gel; and (5) expansion to the fluorination chamber
and mixing with BrF5 (fluorinating reagent) prior to collection
onto the sample-loop silica gel. The latter procedure was used
to test the efficiency with which fluoride products are separated
from analyte oxygen by the GC system and to test the effectiveness of the high-conductance KBr F2 getter in protecting
the downstream silica gel traps. Bromine pentafluoride, which
disassociates to yield small amounts of F2 upon reaction in
vacuum lines, was used as the fluorinating agent for these
experiments because, unlike F2, it could be frozen in to the
sample chamber at low pressures.
All samples were concentrated by cryofocusing at a flow rate
of 3 mL/min. The design of the cryofocus trap permitted
determination of the minimum trapping time by monitoring
m/z 5 32 just after flow reversal. Gas not yet trapped lingers
in the capillary restrictors (Fig. 3) and appears as a broad rise
in background prior to elution of the cryofocused peak. From
this method it was determined that 10 min. is the minimum time
required for complete trapping. Freezing times from the vacuum extraction line to the sample loop trap are discussed
below.
A typical O2 irm-GCMS output is shown in Fig. 4. The first
square-topped peak is the reference gas injected directly into
the source of the mass spectrometer by way of the mixing
volume. Virtually no shift in the voltage ratios (representing
33 32
I/ I and 34 I/ 32 I) is observed for this peak because a bias
was applied so that background ratios approximate those of the
reference gas. The next peak to appear is another aliquot of the
reference O2 gas. In this case the gas was admitted to the
sample loop by flow through the standard-gas capillary, cryofocused onto silica gel, passed through the GC column, and
Isotope-ratio-monitoring of O2
3091
Fig. 4. Mass-specific chromatograms for irm-GCMS on-line run number 737. The d18O value for the Tank 2 reference
gas is 7.3‰ (first peak). The d18O value for the 2.9 nmol sample of Tank 2 O2 injected by cryofocusing is 8.2‰ (second
peak). Correction for blank gives a d18O value of 7.4‰ for the cryofocused peak (Table 2).
finally injected into the mass spectrometer source. Voltage
ratios vary from low to high relative to the reference gas ratio
as the pulse of oxygen passes. The change from low to high
ratios reflects greater adsorption of 33O2 and 34O2 relative to
32
O2 onto the 5A molecular sieve column. This normal isotope
effect, in which the lighter species elute first, arises from
dominance of motions of whole O2 molecules in the adsorbed
state over surface-adsorbate interactions and/or substantial
stiffening of adsorbed O2 intramolecular vibrations caused by
the polar molecular sieve (Van Hook, 1967). It contrasts with
the reverse isotope effect that obtains for CO2 passed through
various separation columns in other irm-GCMS applications.
Efficacy of open tubular columns for separation of different
isotopic molecules of O2 has been documented previously
(Bruner et al., 1966). Integration of the continuously varying
voltage ratios yields 18O/16O and 17O/16O identical (within
0.1‰) to the baseline reference gas ratio after correction for
blank.
The final peak in Fig. 4 arrives approximately 150 sec after
the cryofocused oxygen. Monitoring of m/z during injections
shows this to be a false oxygen signal caused by N2. Virtually
no false signal is evident for m/z 5 32 while the 33 signal is
always three times larger than the 34 signal. Feedback resistances for the 32, 33, and 34 collector amplifiers are 3 3 108V,
3 3 1011V, and 1 3 1011V, respectively. As the sizes of the
false peaks are proportional to the sensitivity of the collectors,
we interpret the spurious peaks as being the result of scattering
of N2 ions toward the collectors. False 18O/16O and 17O/16O
typically yield d17O values of 211‰ and d18O values of 72‰.
As a result, coelution of N2 and O2 can cause significant errors
in oxygen isotope ratio determinations for small samples.
Omission of the column results in large and systematic apparent deviations in d17O and scatter in both d17O and d18O.
Separation of N2 from O2 using the 5A mol sieve column
eliminates excursions in d17O and greatly reduces the scatter in
the data. Purification of analyte O2 by GC after cryofocusing is
necessary for accurate measurements of isotope ratios for sample sizes on the order of several nanomoles.
Analyses of Tanks 1 and 2 obtained with the irm-GCMS
system are summarized in Table 1 and Table 2. These data were
collected after steps were taken to minimize sources of error
discussed below and are indicative of the accuracy and precision of the present technique. Tank 1 proved useful because its
18
O/16O is far removed from that of the blank and so the
influence of the latter was magnified. The 18O/16O of Tank 2 is
more representative of most terrestrial silicate and oxide samples. Results for all tank O2 samples, ranging in size from 2 to
15 nanomoles, are accurate to within 0.1‰ (1s) after correction for blank. Precision is also acceptable, yielding 60.4‰ for
Tank 1 d18O (1 s , n 5 23) and 60.07‰ for Tank 2 d18O (1 s ,
n 5 6). On the basis of clusters of data exhibiting superior
reproducibility, we suspect that the precision for Tank 1 could
be improved with refinements in methods (e.g., improved methods for controlling blank).
System blank was generally on the order of 1 nanomole but
ranged from a low of 0.7 nmol to a high of 1.9 nmol. Blank
peaks are completely obscured by sample peaks, making curve
fitting for blank subtraction of dubious benefit. This poses no
difficulty, however, because the blank is sufficiently constant in
both size and isotopic composition that corrections can be made
by simple mass balance (Fig. 5). Blank from cryofocusing
alone is on the order of 1 nanomole, indicating that the vast
3092
E. D. Young et al.
Table 1. Results of experiments for Tank 1. All values for d18O and d17O are relative to SMOW. Accepted values are 48.0‰ and 24.5‰,
respectively.
Run
763
764
765
766
767
770
771
776
777
788
791
792
810
813
814
816
817
823
824
828
830
831
841
avg4
1
2
3
4
Meas1
d18O
Meas1
d17O
Total
nmol
Blank
d18O
Blank
d17O
Blank
nmol
Corr.2
d18O
Corr.2
d17O
Method
code3
43.46
43.98
44.08
43.92
43.54
46.50
46.61
43.97
43.56
42.98
45.57
45.12
44.76
44.70
45.24
44.58
45.15
42.50
42.42
45.23
42.56
42.25
41.43
22.51
22.51
22.49
22.53
21.89
23.91
24.32
22.25
22.15
21.93
23.34
23.00
22.53
22.19
22.86
22.91
22.7
21.64
21.28
22.87
21.94
21.70
21.38
8.67
9.46
9.40
9.75
9.88
15.40
15.29
8.82
9.40
8.52
13.52
13.27
11.56
11.67
11.19
11.62
11.28
11.71
12.00
11.00
11.81
11.77
10.51
11.79
11.79
11.79
11.79
11.79
8.99
8.99
10.51
10.51
11.69
11.69
11.69
11.29
11.58
10.86
10.86
11.13
11.53
11.53
9.84
10.54
10.54
11.43
5.26
5.26
5.26
5.26
5.26
5.32
5.32
6.43
6.43
5.69
5.69
5.69
4.39
5.49
5.24
5.24
5.44
5.78
5.78
4.19
5.51
5.51
4.74
1.03
1.03
1.03
1.03
1.03
0.72
0.72
0.98
0.98
1.05
1.05
1.05
1.01
1.03
0.91
0.91
0.90
1.81
1.81
0.84
1.60
1.60
1.89
47.73
47.89
48.04
47.70
47.22
48.35
48.48
48.15
47.40
47.39
48.43
48.00
47.97
47.92
48.28
47.44
48.08
48.15
47.90
48.13
47.57
47.24
48.03
47.89
60.37
24.83
24.62
24.61
24.57
23.82
24.83
25.26
24.23
23.98
24.22
24.82
24.49
24.27
24.82
25.42
24.41
24.19
24.54
24.03
24.40
24.51
24.25
25.04
24.44
60.37
SL
SL
SL
SL
SL
SLSG
SLSG
SL
SL
SL
SLSG
SLSG
SLSG
VL
VL
VL
VL
VLC
VLC
SLSG
VLC
VLC
VLC 1 BrF5
Measured values inclusive of blank.
Tank 2 analyses corrected for blank using peak areas in V z sec (areas not shown).
SL 5 collection from sample loop only; SLSG 5 collection from sample loop silica gel; VL 5 collection from vacuum line; VLC 5 collection
from vacuum line and fluorination chamber; VLC 1 BrF5 5 collection from vacuum line and fluorination chamber after mixing with BrF5.
Mean and 1s for all Tank 1 analyses.
majority of extraneous oxygen comes from the He flow system
rather than from the evacuated extraction line.
There is a 0.3‰ downward shift in mean d18O for Tank 1
obtained by expansion to the sample loop without freezing to
silica gel relative to the mean obtained by cryogenic trapping.
The negative shift was accompanied by a deficit of approximately 30% in O2 peak areas relative to trapping experiments
(compare the first eight results in Table 1). Apparently, purging
the sample loop for 10 min (the normal time required for
cryofocusing) is insufficient to flush all of the analyte gas from
the sample loop assembly unless it is first concentrated on to
silica gel.
Adsorption of O2 on to the surfaces of the stainless steel
cryogenic trap (Delchar et al., 1967) located between the fluorination chamber and the sample loop caused a buildup of
oxygen on the trap walls after passage of several samples. As
a result, when the trap was kept cold, blank size was observed
to increase progressively throughout the day until it had more
than doubled. Under these conditions the blank could no longer
be considered constant and the precision of blank-corrected
Table 2. Results of experiments for Tank 2. All values for d18O and d17O are relative to SMOW. Accepted values are 7.3‰ and 3.4‰, respectively.
Run
750
738
737
731
746
000
avg4
1
2
3
4
5
Meas1
d18O
Meas1
d17O
Total
nmol
Blank
d18O
Blank
d17O
Blank
nmol
Corr.2
d18O
Corr.2
d17O
Method
code3
7.71
8.28
8.17
8.06
7.64
7.74
3.66
3.86
4.08
3.30
3.43
3.41
8.54
2.54
2.87
2.91
6.40
5.19
10.67
10.05
10.05
10.05
10.05
10.05
6.01
4.755
4.755
4.755
4.755
4.755
0.92
0.835
0.835
0.835
0.835
0.835
7.35
7.45
7.43
7.29
7.29
7.31
7.35
60.07
3.38
3.43
3.81
2.73
3.23
3.16
3.29
60.36
SL
SL
SL
SL
SL
SL
Measured values inclusive of blank.
Tank 4 analyses corrected for blank using peak areas in V z sec (areas not shown).
SL 5 collection from sample loop.
Mean and 1s for all Tank 2 analyses.
Average blank over period of analyses.
Isotope-ratio-monitoring of O2
3093
purification by gas chromatography to preclude false signals
from interfering ions and periodic cleaning of cryogenic traps
to remove adsorbed oxygen. When combined with laser ablation and fluorination, irm-GCMS can be used to obtain in situ
high-precision oxygen isotope ratio analyses of microgram
quantities of silicate minerals.
Acknowledgments—The first author wishes to thank the Carnegie Institution of Washington’s Geophysical Laboratory for support as a
postdoctoral fellow during the early stages of this study. Professor
R. N. Clayton kindly shared with us his experiences regarding calibration of 17O/16O. Thoughtful reviews by Professor J. W. Valley and Dr.
Z. D. Sharp are greatly appreciated.
Fig. 5. Plot of measured d18O and sample size for on-line irm-GCMS
analyses of Tank 2 prior to blank correction (solid circles). Grey line is
the hyperbolic best fit to the data. Mixing between a single blank and
Tank 2 would yield a perfect hyperbolic curve. The correlation coefficient of 0.969 for this fit is indication that the blank for these runs was
effectively constant in both size and d18O value. Cross shows typical
measured blank values.
results declined. The problem was solved by heating the trap to
423 K between samples.
5.2. Effects of fluorination
Results of experiments on O2 test gases have been used to
develop an oxygen isotope microprobe based on UV laser
ablation and fluorination. The details of the UV laser ablationfluorination system connected to the on-line system at Oxford
University are presented elsewhere (Young et al., 1998). We
mention here, however, that fluorination experiments revealed
that the use of a separation column simplifies measurement of
d17O.
Gaseous fluorides can adversely affect the accuracy and
precision of d17O measurements (Clayton and Mayeda, 1983).
These fluorides are present as products of fluorination by either
BrF5 or F2. Analysis by conventional MS requires a purification
procedure involving cryogenic separation of O2 from potential
fluoride contaminants, especially NF3. In combination with
laser ablation-fluorination (using purified F2), the irm-GCMS
method described here yields d17O and d18O values of terrestrial samples relative to SMOW that are consistent with the
average terrestrial mass fractionation line within 1s uncertainties without additional purification of the analyte gas. Similarly,
analyses of meteoritical samples lying off the terrestrial line are
analyzed with accuracy in both d18O and d17O (Young et al.,
1998).
6. CONCLUSIONS
The experiments described here show that irm-GCMS can be
used to analyze nanomole quantities of O2 gas for 18O/16O and
17
O/16O with accuracy and precision suitable for geochemical
applications, including high-temperature studies in which variations in d18O of ,1‰ are important and investigations of
meteoritical materials where relations between d18O and d17O
are crucial. Critical aspects of the analytical method are gas
REFERENCES
Bigeleisen J. (1965) Chemistry of isotopes. Science 147, 463– 471.
Bruner F., Cartoni G. P., and Liberti A. (1966) Gas chromatography of
isotopic molecules on open tubular columns. Anal. Chem. 38, 298 –
303.
Clayton R. N. and Mayeda T. K. (1983) Oxygen isotopes in eucrites,
shergottites, nakhlites, and chassignites. Earth Planet. Sci. Lett. 62,
1– 6.
Craig H. (1972) Atmospheric oxygen: Isotopic composition and solubility fractionation. Science 175, 54 –55.
Delchar T. A. and Tompkins F. C. (1967) Chemisorption and
incorporation of oxygen at a nickel surface. Proc. Royal Soc. 300,
141–158.
Hayes J. M., DesMarais D. J., Peterson D. W., Schoeller D. A., and
Taylor S. P. (1977) High precision stable isotope ratios from microgram samples. Adv. Mass Spectrom. 7A, 475– 480.
Jones A. M., Iacumin P., and Young E. D. (1998) High-resolution d18O
analysis of biogenic phosphate by GC-irms and UV laser fluorination. Chem. Geol. (submitted).
Klemp M. A., Akard M. L., and Sacks R. D. (1993) Crofocusing inlet
with reverse flow sample collection for gas chromatography. Anal.
Chem. 65, 2516 –2521.
Matsuhisa Y., Goldsmith J. R., and Clayton R. N. (1978) Mechanisms
of hydrothermal crystallization of quartz at 250°C and 15 kbar.
Geochim. Cosmochim. Acta 42, 173–182.
McKinney C. R., McCrea J. M., Epstein S., Allen H. A., and Urey H. C.
(1950) Improvements in mass spectrometers for the measurement of
small differences in isotope abundance ratios. Rev. Sci. Instr. 21,
724.
Merritt D. A. and Hayes J. M. (1994) Factors controlling precision and
accuracy in isotope-ratio-monitoring mass spectrometry. Anal.
Chem. 66, 2336 –2347.
Merritt D. A., Brand W. A., and Hayes J. M. (1994) Isotope-ratio
monitoring gas chromatography-mass spectrometry: methods for
isotopic calibration. Org. Geochem. 21, 573–583.
Nier A. O. (1950) A redetermination of the relative abundances of the
isotopes of carbon, nitrogen, oxygen, argon, and potassium. Phys.
Rev. 77, 789 –793.
Rumble D. III, Farquhar J., Young E. D., and Christensen C. P. (1997)
In situ oxygen isotope analysis with an excimer laser using F2 and
BrF5 reagents and O2 gas as analyte. Geochim. Cosmochim. Acta 61,
4229 – 4234.
Santrock J., Studley S. A., and Hayes J. M. (1985) Isotopic analyses
based on the mass spectrum of carbon dioxide. Anal. Chem. 57,
1444 –1448.
Valley J. W., Kitchen N., Kohn M. J., Niendorf C. R., and Spicuzza M.
(1995) UWG-2, a garnet standard for oxygen isotope ratios: strategies for high precision and accuracy with laser heating. Geochim.
Cosmochim. Acta 59, 5223–5231.
Van Hook W. A. (1967) Isotope effects on vaporization from the
adsorbed state. The methane system. J. Phys. Chem. 71, 3270 –
3275.
3094
E. D. Young et al.
Weichert U. and Hoefs J. (1995) An excimer laser-based micro analytical preparation technique for in-situ oxygen isotope analysis of
silicate and oxide minerals. Geochim. Cosmochim. Acta 59, 4093–
4101.
Young E. D., Coutts D. W., and D. Kapitan (1998) UV laser ablation
and irm-GCMS microanalysis of 18O/16O and 17O/16O with application to a calcium-aluminium-rich inclusion from the Allende meteorite. Geochim. Cosmochim. Acta 62, 3161–3168 (this issue).